CPU / GPU Performance Per Price

data sources
GPU benchmarks: https://www.userbenchmark.com/page/developer
CPU benchmarks: https://browser.geekbench.com/opencl-benchmarks
GPU and CPU prices: https://pcpartpicker.com/

library(shiny)
library(tidyverse)
library(plotly)
library(ggrepel)

gfx  <- read_csv("GPU_UserBenchmarks.csv")
pcpp <- read_csv("pcpartpicker.com-(Crawl-Run)---2020-07-20T010546Z.csv")

gfx$model_tidy <- NA
pcpp <- pcpp %>% 
  mutate(Chipset_mod = gsub("GeForce |Radeon ", "", Chipset))

for (i in unique(pcpp$Chipset_mod)) {
  row_hits <- grep(paste0("\\b", i, "\\b"), 
                   gfx$Model, 
                   ignore.case = TRUE)
  if (length(row_hits) > 0) {
    gfx$model_tidy[row_hits] <- 
      c(rep(i, length(row_hits)))
  }
}

left_join(gfx, pcpp, by = c("model_tidy" = "Chipset_mod")) %>% 
  select(model_tidy, Benchmark, Price, Samples) %>% 
  group_by(model_tidy) %>% distinct() %>% 
  arrange(desc(model_tidy, Price)) %>% slice(n()) %>% drop_na() %>% 
  ungroup() %>% mutate(Price = as.numeric(gsub("[$]|\\..*", "", Price))) %>%
  mutate(`Score/dollar` = Benchmark/Price) %>%
  ggplot(aes(x     = Benchmark, 
             y     = `Score/dollar`, 
             color = Benchmark,
             size  = Samples)) + 
  geom_point() + 
  geom_text_repel(
    aes(label = ifelse(Benchmark > 50 | `Score/dollar` > 0.15, 
                       paste0(model_tidy, " (", Price, " USD)"),
                       '')), 
    hjust = 0, vjust = 0,
    size = 3) + 
  scale_color_gradient(low = "black", high = "red") + 
  guides(color = FALSE) + 
  labs(x        = "UserBenchmark Score", 
       y        = "Score / Dollar",
       title    = "GPU Performance Per Dollar",
       subtitle = "Prices from PC Partpicker (July 2020)") +
  scale_x_continuous(trans = "log10") + 
  theme_light()

cpu_geek <- read_csv("browser.geekbench.com-(Crawl-Run)---2020-07-20T044700Z.csv")
cpu_pcpp <- read_csv("pcpartpicker.com-(Crawl-Run)---2020-07-20T045353Z.csv")

cpu_geek <- cpu_geek %>% mutate(CPU = toupper(Processor))
cpu_pcpp <- cpu_pcpp %>% mutate(CPU = toupper(`Name Wrapper`))

cpus <- left_join(cpu_geek, cpu_pcpp, by = "CPU") %>% 
    select(CPU, Score, Price, `Speed (Cores)`) %>% drop_na() %>% 
    group_by(CPU) %>% arrange(desc(CPU, Score)) %>% slice(1) %>% 
    ungroup() %>% mutate(Price = as.numeric(gsub("[$]|\\..*", "", Price))) %>% 
    mutate(`Score/Dollar` = Score/Price,
           CPU = str_to_title(CPU)) %>% 
    separate(`Speed (Cores)`, 
             sep = " GHz ", 
             into = c("Speed (GHz)", "Cores")) %>% 
    mutate(Cores = as.numeric(gsub("\\(| cores\\)", "", Cores)), 
           `Speed (GHz)` = as.numeric(`Speed (GHz)`),
           Processor = paste0(CPU, " (", `Speed (GHz)`, " GHz)"))

cpus %>%
  filter(Cores >= 8,
         `Speed (GHz)` >= 3) %>%
  ggplot(aes(x     = reorder(Processor, `Score/Dollar`), 
             y     = `Score/Dollar`, 
             fill  = Score)) + 
  geom_bar(stat = "identity") +
  guides(color = FALSE) + 
  labs(x        = "Processor", 
       y        = "Geekbench Score / Dollar",
       title    = "CPU Performance Per Dollar (Single Core, 8+ Cores)",
       subtitle = "Prices from PC Partpicker (July 2020)") +
  coord_flip() +
  theme_light()

cpus %>%
  filter(Cores < 8,
         `Score/Dollar` >= 5,
         `Speed (GHz)` >= 3) %>%
  ggplot(aes(x     = reorder(Processor, `Score/Dollar`), 
             y     = `Score/Dollar`, 
             fill  = Score)) + 
  geom_bar(stat = "identity") +
  guides(color = FALSE) + 
  labs(x        = "Processor", 
       y        = "Geekbench Score / Dollar",
       title    = "CPU Performance Per Dollar (Single Core, < 8 Cores)",
       subtitle = "Prices from PC Partpicker (July 2020)") +
  coord_flip() +
  theme_light()

(cpus %>%
  filter(`Speed (GHz)` >= 3.3,
         Cores >= 4) %>%
  ggplot(aes(x     = Score,
             y     = `Speed (GHz)`,
             color = `Score/Dollar`,
             label = CPU)) +
  geom_point() +
  labs(x = "Geekbench Score",
       y = "Speed (GHz)") +
  scale_color_gradient(low = "black", high = "red") +
  theme_light()) %>%
  ggplotly() %>%
  layout(title = list(text = paste0('CPU Performance Per Dollar (Single Core, 4+ Cores)',
                                    '<br>',
                                    '<sup>',
                                    'Prices from PC Partpicker (July 2020)',
                                    '</sup>')),
         margin = 1)